diin papr riftp.iza.org › dp11515.pdf · is no proven evidence on long-term behavior after being...
TRANSCRIPT
DISCUSSION PAPER SERIES
IZA DP No. 11515
Horst EntorfJia Hou
Financial Education for the Disadvantaged? A Review
APRIL 2018
Any opinions expressed in this paper are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but IZA takes no institutional policy positions. The IZA research network is committed to the IZA Guiding Principles of Research Integrity.The IZA Institute of Labor Economics is an independent economic research institute that conducts research in labor economics and offers evidence-based policy advice on labor market issues. Supported by the Deutsche Post Foundation, IZA runs the world’s largest network of economists, whose research aims to provide answers to the global labor market challenges of our time. Our key objective is to build bridges between academic research, policymakers and society.IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion. Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
Schaumburg-Lippe-Straße 5–953113 Bonn, Germany
Phone: +49-228-3894-0Email: [email protected] www.iza.org
IZA – Institute of Labor Economics
DISCUSSION PAPER SERIES
IZA DP No. 11515
Financial Education for the Disadvantaged? A Review
APRIL 2018
Horst EntorfGoethe University Frankfurt, SAFE and IZA
Jia HouGoethe University Frankfurt and SAFE
ABSTRACT
IZA DP No. 11515 APRIL 2018
Financial Education for the Disadvantaged? A Review*
In contrast to the popularity of financial education interventions worldwide, studies on
the economic effects of those interventions report mixed results. With a focus on the
effect on disadvantaged groups, we review both the theoretical and empirical findings
in order to understand why this discrepancy exists. The survey first highlights that it is
necessary to distinguish between the concepts of, and the relationships between, financial
education, financial literacy and financial behavior to identify the true effects of financial
education. The review addresses possible biases caused by third factors such as numeracy.
Next, we review theories on financial literacy which make clear that the effect of financial
education interventions is heterogeneous across the population. Last, we look closely at
main empirical studies on financial education targeted at the migrants/immigrants, the
low-income earners and the young, and compare their methodologies. There seems to be
a positive effect on short-term financial knowledge and awareness of the young, but there
is no proven evidence on long-term behavior after being grown up. Studies on financial
behavior of migrants and immigrants show almost no effect of financial education.
JEL Classification: G28, G41, I24, I25, I28
Keywords: financial education, financial literacy, inequality, program, evaluation
Corresponding author:Horst EntorfGoethe University FrankfurtFaculty of Economics and Business AdministrationTheodor-W.-Adorno-Platz 460323 Frankfurt am MainGermany
E-mail: [email protected]
* We are grateful to the LOEWE Center Sustainable Architecture for Finance in Europe (SAFE) program for financial
support of our research.
IZA DP No. 11515 APRIL 2018
NON-TECHNICAL SUMMARYMany researchers and politicians have identified a lack of financial literacy among the
disadvantaged groups of the society. Thus, unequal access to financial markets seems
to be a major source of economic inequality, and tackling inequality by starting financial
education initiatives seems to be a straightforward solution. However, in contrast to the
popularity of financial education interventions worldwide, studies on the economic effects
of those interventions report mixed results.
With a focus on the effect of disadvantaged groups, we review both the theoretical and
empirical findings in order to understand why this discrepancy exists. The survey highlights
that it is necessary to distinguish between the concepts of financial education, financial
literacy and financial behavior in order to disentangle cause and effect. To understand the
heterogeneity of potential outcomes, we review theoretical explanations which address the
effect of interventions across the population and over the life cycle. Looking at empirical
results, we discuss the importance of potentially neglected factors such as cognitive ability,
mathematical capability or numeracy, and we find that the young, the immigrants/migrants
and the low-income groups obtain more attention from scholars than other disadvantaged
groups.
Some support in favor of financial education for the young has been detected in terms of
higher short-term financial knowledge and awareness, but there is no proven evidence of
improved long-term behavior after growing up. From the methodological point of view, in
recent years randomized control trials (RCT) have become a popular method to evaluate the
effects of financial education, in particular for studies on remittance behavior of migrants.
Our review discusses some limitations of both RCTs and econometric studies which might
be taken into account by future research.
In sum, despite the enthusiasm of many politicians, who see financial education and financial
literacy as the key to tackle the problem of financial vulnerability and economic inequality,
so far, there is no clear evidence, or at least no scientific consensus, on the effectiveness
of performed interventions. The constant search for effective financial education might
even cost enormously such that the costs of financial education programs would outweigh
potential benefits. It is perhaps true that the financial behavior of the poor is arguably
more controlled by lack of aspirations such that financial policy should be directed towards
behaviorally motivated anti-poverty policies or simply education in general.
1 Introduction
In the last two decades, financial education has gained much attention from policy
makers around the world. As noted by OECD (2015b), 59 countries are reported to
be developing a national strategy for financial education by 2015. For instance, fi-
nancial education has been compulsory on the national curriculum in England since
2014, after a petition with more than 100,000 signatures was submitted for the de-
bate in 2011. In the U.S., a “Financial Literacy and Education Improvement Act”
was part of the “Fair and Accurate Credit Transactions Act” of 2003 to develop
a national financial education strategy. In the European Union, in its 2008 reso-
lution on improving consumer education and awareness on credit and finance, the
European Parliament noted that raising the level of financial literacy of consumers
should be a priority for policy-makers, both at the Member State and the European
level1.
Studies on financial literacy have confirmed the importance of financial knowl-
edge on the individual’s personal finance2 and economic outcomes3. The worldwide
2008 financial crisis reveals that individuals’ ability to face the crisis is different
according to their financial literacy (for example, Klapper et al. (2013)). It is then
intuitive to think that financial education is critical to individuals’ welfare or to
protect them from economic crisis, since financial education is supposed to improve
the financial knowledge of the individual and eventually to improve the resilience
of the economy. If so, does financial education affect financial behavior through
the channel of financial literacy? Are financial knowledge and, more importantly,
financial behavior, really affected by financial education, or is it more reasonable to
assume that third factors such as numeracy of mathematical capabilities play the
leading role? Why are the empirical findings on the effects of financial education so
mixed while financial education interventions remain so popular? Is it possible that
financial education policy is just an excuse and loophole for politicians who do not
want to forego any opportunity to avoid being accused of any future financial crisis,
1http://www.europarl.europa.eu/sides/getDoc.do?type=TA&language=EN&reference=
P6-TA-2008-539.
2For example, Hastings et al. (2013) list behavior related to credit card spending, holdingstocks, bonds, mutual funds or other securities, loans (e.g., making late payments on a mortgage,comparison shopping for a mortgage or auto loan), insurance coverage, financial counselling (e.g.,seeking professional advice about a mortgage, loan, insurance, tax planning, or debt counselling).A different focus includes impacts on portfolio choice (Campbell (2006), Von Gaudecker (2015))and saving and borrowing (Gathergood (2012)).
3For example,Van Rooij et al. (2011) on stock market participation and Disney and Gathergood(2013) on consumer portfolios. Please see Lusardi and Mitchell (2014) and Hastings et al. (2013)for a more extensive review.
2
whatever the (unknown) costs4 of such strategy? This survey aims at answering
these questions by reviewing the current literature.
Our contribution is threefold. Firstly, we distinguish between the concepts of
financial education, financial literacy (or financial knowledge5, respectively), and
financial behavior, and we define and consider the three concepts of being in a
logical order. We start by highlighting the link between financial education inter-
ventions and measured financial knowledge/literacy and go from there to (good or
bad) financial decision making of individuals and households. We discuss research
results which lead to question the causal influence of financial education on behavior,
because other factors such as mathematical ability might be likewise important, ren-
dering the influence of financial education on outcomes as a potentially spurious one.
Secondly, this survey is the first to review theories on financial education. The-
ories on financial literacy, which incorporate the accumulation process of financial
literacy, are the most relevant ones. The human capital accumulation model and
endogenous investment model of financial literacy imply that different groups of
people are in heterogeneous need of financial education.
Thirdly, we take account of the heterogeneity of financial education interven-
tions and provide a review targeted at disadvantaged groups. As recognized by the
OECD, young people (more than half of the responding countries in the OECD sur-
vey), women, low-income groups, elderly people, micro-, small-, and medium-sized
enterprises, migrants and, in a few countries, people living in rural areas, are tar-
geted groups of importance for financial education strategies (OECD (2015b)).
Different from comprehensive meta-analysis by Fernandes et al. (2014), Miller
et al. (2015) and Kaiser and Menkhoff (2017), we review research papers on a study-
by-study basis in order to focus on those studies which allow us to gain some knowl-
edge on the impact of financial education on groups at the bottom of financial
inequality. Moreover, we analyze methodological discrepancies, which are often re-
lated to the question whether randomized control trials (RCT) are applied or not.
We find that the young, the immigrants/migrants and the low-income groups ob-
tain more attention from scholars than other groups. Studies on large-scale samples,
both using RCTs or econometric methods for quasi-experiments, show up more of-
4Willis (2011) criticizes the over-support from politics on for financial education and its ”time,money, privacy, and autonomy costs”.
5As in Lusardi and Mitchell (2014), we use the expression “financial knowledge” as a synonym.
3
ten for investigating financial education of the young. For the immigrants/migrants,
the effect on remittance behavior is the focus. Besides, it is worth of noting that
each RCT or quasi-experiment we study is faced with some limitations, which we
would like to discuss and which might be taken into account by future studies.
The survey is organized as follows. Section 2 discusses the link between fi-
nancial education, financial literacy and financial behavior, and draws attention to
the importance of third-factor influences. Section 3 summarizes theoretical foun-
dations of financial literacy. Section 4 reviews the empirical findings regarding the
influence of financial education interventions on disadvantaged groups, mainly the
immigrants/migrants, the low-income groups and the young. Section 5 summarizes
the survey and offers some conclusions.
2 Financial Behavior: Financial Education, Fi-
nancial Literacy, Numeracy or Just Education?
The literature of interest is split between evaluating the effects of financial ed-
ucation on either the targeted financial behavior or on financial knowledge. For
example, the criticism on policy-driven financial education by Willis (2011) starts
with the argument that ”objective observers generally admit that research to date
does not demonstrate a causal chain from financial education to higher financial lit-
eracy to better financial behavior to improved financial outcomes”, which implicitly
defines that an effective financial education should be an intervention able to even-
tually improve financial behavior. However, Fernandes et al. (2014) refer to financial
education interventions directly as ”manipulated financial literacy”. Unclear inten-
tions of financial education might blur findings from the literature. We would like
to avoid such conceptual ambiguity and therefore consider financial literacy (and
financial knowledge) as a means to an end, but not the end itself. Nevertheless,
“financial literacy”, per se, plays the more glamorous role in many research articles
such that a review of the literature would be incomplete without them.
2.1 Financial Education and Financial Literacy
Actually, the definition of financial literacy has evolved over time (Huston (2010)),
but still with no convergence. Lusardi and Mitchell (2014), perhaps the most promi-
nent contributors to the field of study, refer to financial literacy as “peoples’ ability
to process economic information and make informed decisions about financial plan-
4
ning, wealth accumulation, debt, and pensions”. Financial literacy can then be used
interchangeably with “financial knowledge”, which affects individuals’ financial be-
havior or economic outcomes. Each individual is motivated differently to accumulate
her/his financial literacy based on the cost and return of financial literacy over the
life cycle. In this context, the effects of financial education are suggested to be
heterogeneous by age and income level, etc. The OECD, one of the most active or-
ganizations in promoting the awareness of financial literacy and financial education,
use a more target-based concept by directly focusing on the purpose of financial
literacy, i.e., “financial inclusion”. Their proposed measurement “toolkit” incorpo-
rates capturing “. . . information about financial behavior, attitudes and knowledge,
in order to assess levels of financial literacy and financial inclusion” (OECD (2015a)).
Most literature reveals a strong link between financial literacy and poor or smart
financial decision making, whereas the causal influence of financial education on fi-
nancial knowledge or even financial decision making is less clear, as has been docu-
mented in detail by Hastings et al. (2013), Fernandes et al. (2014) and Lusardi and
Mitchell (2014), among many others. There are several reasons for contradictory
assessments of the effects of financial education. For instance, studies are usu-
ally based on different target groups or different methodological approaches, where
the use or nonuse of random control trials (RCT) is of increasing importance (see
Hastings et al. (2013), Lusardi and Mitchell (2014), or Kaiser and Menkhoff (2017)).
A substantial part of the debate on the effectiveness of financial education is
linked to the way how researchers define and measure financial literacy itself. Hung
et al. (2009) document the breadth of existing conceptual definitions, ranging from
financial knowledge and familiarity with financial terms to the capability of making
informed judgements. As also noticed by Lusardi and Mitchell (2014) and Xu and
Zia (2012), many studies follow Hilgert et al. (2003), Lusardi and Mitchell (2008),
Lusardi et al. (2010) etc., and measure financial literacy with respondents’ correct
answer to the ”big three” questions on inflation, interest rate and risk aversion.
With the efforts of those studies, more national surveys build on those questions
when measuring financial literacy of households6 (e.g., the Panel on Household Fi-
nance of Germany, the Health and Retirement Study and National Financial Capa-
bility Study of the U.S.). Additional financial literacy questions like questions on
mortgages and bond pricing are also included in some studies.
However, how valid those questions and correct answers are for capturing the
6Please see Lusardi and Mitchell (2011), Hastings et al. (2013), Xu and Zia (2012) for compar-isons of financial literacy measured by the ”big three” around the world.
5
true financial knowledge is still under debate, letting alone the variation caused by
different methods of obtaining the information (such as online surveys or telephone
interviews) or ways of rating the knowledge (e.g., Huston (2010)). Schuhen and
Schurkmann (2014) indicate that the reason why Germany refused to participate in
the OECD-PISA assessment of financial literacy7 is that “the current version of a
Financial Literacy assessment is not sufficiently developed and reliable findings are
therefore not possible”.
2.2 Neglected Factors in the Analysis of Financial Behavior
The work by Willis (2008, 2011) has gained much attention for its harsh criticism
of financial education policy and for casting serious doubt on the widespread belief
in the effectiveness of financial education (she also employs the combined term “fi-
nancial literacy education”). She argues that the gulf between literacy and useful
decisions on financial products cannot realistically be bridged, for instance, because
new products are often highly complex and changing over time such that financial
literacy education would mean chasing a moving target it would never reach. On the
contrary, outdated lessons or financial rules of thumb may not only be irrelevant, but
even counterproductive. She refers to the problem that financial education appears
to increase confidence without improving ability, i.e., overconfidence, potentially
leading to worse decisions. This is in line with recent research by Von Gaudecker
(2015), who finds that most losses from lacking diversification are incurred by over-
confident investors. McCannon et al. (2016) show that overconfident clients, who
believe they possess more knowledge and understanding than they actually do, lead
investors to underestimate risks and shortcomings of certain investment options, al-
though results by Hackethal et al. (2012) suggest that such biased behavior seems
to diminish with experience.
Other researchers identify potentially omitted variables when linking financial
literacy, financial education, and financial decision making. A recent paper by
Gramat,ki (2017) suggests that measured financial literacy could just reflect math-
ematical capacity. Using the first OECD PISA (2012) international assessment of
financial literacy, Gramat,ki (2017) finds that the financial literacy gap between na-
tive and immigrant students is significant, but not after controlling for students’
math score. Hung et al. (2009) point out that many concepts, such as the ability to
work with numbers, i.e., numeracy, share features with financial literacy. They also
7The Program for International Student Assessment (PISA) is an international project con-ducted by OECD to analyze the financial literacy of adolescents aged 15-16 in the member states.In 2012 and 2015, 18 countries and 16 countries participated in the assessment, but not Germany.
6
describe other factors such as general knowledge, cognitive abilities and decision-
making competence and their relationship to financial literacy. The skill captured
by the standard financial literacy question on, say, compound interest might just be
an indicator of general numeracy skills rather than financial literacy. After including
numeracy, the capability of long-term planning, willingness to take prudent invest-
ment risks, and confidence with respect to marketplace decisions, Fernandes et al.
(2014) show that financial literacy ceases to be significant in most of the presented
regressions. The authors correctly mention different interpretations of their results.
The argument against financial literacy would be that third factors cause financial
behavior, such that financial literacy would only be seemingly related to good or
bad financial decisions. However, their results do not preclude the opposite inter-
pretation that included covariates are endogenous and caused by financial literacy.
The obvious question is what matters most for good financial decision making
and behavior. Perhaps it is just education? Cole et al. (2014) study the causal
effects of education on asset accumulation and financial market participation, where
they use changes in state compulsory education laws as an instrument for educa-
tional attainment. Their results are also in line with the hypothesis that the link
between education and financial outcomes is indirect, as education improves cogni-
tive ability (see Hanushek and Woessmann (2008)) and cognitive ability appears to
improve financial outcomes. Thus, despite the strong link between education and
financial literacy (Lusardi and Mitchell (2014)), the correlation between financial
literacy and financial outcomes might not be a causal one, but caused by the joint
dependence on education, particularly in maths skills. This calls into question the
effectiveness and efficiency of (costly) financial education policies.
However, before jumping to conclusions, we should have a better theoretical
understanding of the effects of financial education interventions. Moreover, financial
education programs can be tailored to the specific needs of target groups such that
there might be group-specific outcomes and simple one-fits-all conclusions might not
exist.
3 Theoretical Foundations of Effects of Financial
Education Interventions
Financial education can be seen as a particular type of education which can be
subsumed under the theory of human capital formation. Here, financial education is
usually seen as a driver of financial literacy (which also means financial knowledge,
7
as defined by Lusardi and Mitchell (2014)).
3.1 Financial Education to Accumulate Financial Literacyas Human Capital
Delavande et al. (2008) propose a two period life cycle model of financial knowledge
on retirement saving and portfolio choice. Departing from the model of Kezdi and
Willis (2008), which is due to the optimal portfolio choice model of Merton (1969),
Delavande et al. (2008) assume that there is heterogeneity in people’s knowledge
about financial markets and thus not all of them can construct the optimal portfolio
choices. More financial knowledge enables the individual to make better financial
decisions. The accumulation of financial knowledge follows the human capital pro-
duction function proposed by Ben-Porath (1967) and later developed by Cunha and
Heckman (2007).
In the model of Delavande et al. (2008), financial knowledge is accumulated in a
Cobb-Douglas way8:
∆ft =dftdt
= α(etft)β1Hβ2
t Mβ3t E
β4t ; β1 + β2 + β3 + β4 < 1, (1)
where ∆ft is the financial knowledge obtained in the period t, et is effort devoted to
learning, ft is the stock of financial knowledge at the beginning of a period, and etft
is the number of efficiency units in time invested in obtaining financial knowledge.
With this production model, financial knowledge is self-productive and the efficiency
of obtaining financial knowledge in the future can be enhanced by disposing of higher
current financial knowledge stock. Besides, the productivity of investment in learn-
ing financial knowledge is dependent on other factors captured by Ht,Mt, Et, such
as the stock of other human capital, the knowledge of family and friends etc.
The model implies that those people who have lower inputs in Ht,Mt, Et need
more help from public financial education interventions since they need more time
in obtaining the same financial knowledge. However, the model does not say too
much about the efficiency of financial education interventions in different institu-
tional settings, such as more or less generous social security systems. Besides, it is
not clear yet what is in Ht,Mt, Et.
In order to explain the heterogeneity of beliefs observed by the American Health
8This is equation (4) in Delavande et al. (2008). We change some of the notations for theconsistency of this survey, similarly for the following equations.
8
and Retirement Study about the stock market, Kezdi and Willis (2011) propose
a three-period life-cycle model of financial knowledge. The model suggests that
those who have higher lifetime earnings, higher risk tolerance, and more patience
are those who are more motivated to obtain financial knowledge. The implication
on financial education interventions is two-fold. On one hand, financial education
interventions will be more attractive to those who are more motivated to obtain
financial knowledge. On the other hand, those who are less motivated to obtain
financial knowledge need more exogenous financial education interventions. The
model of Kezdi and Willis (2011) can be seen as an application of Delavande et al.
(2008).
3.2 Financial Education to Accumulate Financial Literacyas Endogenous Choice Variable
In a framework different from the one incorporating financial literacy into human
capital accumulation, Jappelli and Padula (2013) propose a consumption model
of consumer investment in financial literacy based on Arrow (1987). In the model,
financial literacy can increase the return on wealth, but also costs in terms of current
consumption and it decays over time. In the multi-period case, financial literacy
evolves as the following9:
(i− α stft+1
)fαt+1 − i(1− δ) = 0 for t ≤ T − 3 (2)
i− α stft+1
= 0 for t = T − 2, (3)
where i is the cost of financial literacy in terms of consumption good, st is the saving
at period t. α is the return on the stock of financial literacy, ft+1. The financial
literacy depreciates at the rate of δ.
The model implies that both financial literacy and wealth are endogenous and
strongly correlated with each other over the life cycle. An interesting prediction
of the model of Jappelli and Padula (2013) is that in countries with a more gener-
ous social security system, investors will be less motivated to accumulate financial
literacy. This feature hints at an interesting source of international heterogeneity of
financial literacy. It further indicates that the need for financial education interven-
tions might depend on country characteristics such as the prevailing social security
system.
9These are equation (8) and (9) in Jappelli and Padula (2013).
9
More recently, Lusardi et al. (2017) develop a stochastic life cycle model of fi-
nancial knowledge. The model follows the line of Jappelli and Padula (2013), but
extends it by incorporating many features such as borrowing constraints, mortal-
ity risk, demographic factors, stock market returns, and earnings and health shocks,
which brings it closer to the standard model of saving (Lusardi and Mitchell (2014)).
As in Jappelli and Padula (2013), the model is based on the assumption that finan-
cial knowledge is acquired endogenously over the life cycle.
The model is described by a intertemporal value function, which is subject to
borrowing constraints. Financial literacy at period t+1, ft+1, costs (it) and decays
(δ):
ft+1 = (1− δ)ft + it, (4)
but it increases the stochastic return on risky assets. The model is solved by numer-
ically by using backward recursion method. To calibrate the effects of endogenous
financial knowledge on the wealth inequality, Lusardi et al. (2017) impose value on
parameters based on previous studies including Hubbard et al. (1995), Attanasio
(1999) and Scholz et al. (2006).
Several predications based on different sources of uncertainty can be drawn from
the model. An important one is that financial education interventions affect sub-
groups of the population differently. Participants who find it not optimal to save
more, such as the low educated and low-income group, might cause the ineffective-
ness of financial education interventions targeted at improving saving. It is then
necessary to design special financial education interventions in a way that increases
the participation rate of those groups.
To extend the standard two-period model, Lusardi et al. (2017) take the follow-
ing steps: (1) introducing uncertainty regarding asset returns, household income,
and out-of-pocket medical expenditures; (2) including stochastic mortality risk for
individuals; (3) examining different education groups. The simulation study based
on various settings offers a large variety of implications which rationalize some of
the large differences in wealth found in much prior empirical works on saving. The
results also show that some level of financial ignorance may actually be optimal.
For those who cannot benefit from the financial knowledge and greater financial
sophistication, costs of acquiring financial education will exceed its benefits.
Overall, theories of financial literacy illustrate its accumulation process and pro-
vide a potential leverage point for improving the efficiency of financial education
10
interventions. The model of Delavande et al. (2008) incorporates financial literacy
into a human capital accumulation function. In their framework, financial literacy
is exogenous and determined by investment and many other factors such as social
network and cognitive ability. In contrast, Jappelli and Padula (2013) and Lusardi
et al. (2017) model wealth and financial literacy as endogenous choice variables. The
open question is what are the crucial third factors driving the endogenous cycle.
Given the financial behavior of agents is well-understood and fully covered, the
models have much power in explaining the saving behavior and wealth. Endogeneity
of financial knowledge certainly is a realistic assumption, but effective implementa-
tion of financial education seems to require the use of “teachable moments” which
would imply some ”just-in-time education”. For example, after receiving a bequest,
when taking out a mortgage, when figuring out when to retire, or when graduate
students realize the importance of financial knowledge after graduating and receiv-
ing the first paycheck, income and capital earners might become very motivated to
obtain knowledge. Of course, theoretical modeling does not deliver the identification
of such turning points of life, the analysis and evaluation of those “trigger points”
remains an empirical question. However, theory helps us to understand the short-
and long-run and highly heterogeneous consequences of interventions on wealth ac-
cumulation.
Willis (2008) is even skeptical that offering financial education at a teachable
moment, which also needs to be a “reachable” moment, has a lasting effect and that
people are more likely to learn about personal finance. Empirical evidence from a
large meta study by Kaiser and Menkhoff (2017) suggests that offering financial ed-
ucation at a teachable moment indeed shows significant overall effects, but teaching
low-income groups has less impact. Thus, effects of financial education seem to be
subject to target-group influence.
4 Quantifying the Effects of Financial Education
Compared with the rather limited theoretical work, there is an increasing number
of empirical studies on the effects of financial education. Using Google Search, the
search for “financial education empirical research paper pdf” delivers about 43,200
hits (01 April 2018). Not surprisingly, the existing research delivers a large data
base for recent meta-studies on financial education which we are going to survey in
the next subsection. We also focus on (treatment) effects for disadvantaged groups
and we particularly draw the attention to differences between random control trials
11
(RCT) and non-RCT studies.
4.1 Evidence from Meta Studies
Two comprehensive meta studies, Fernandes et al. (2014) and Kaiser and Menkhoff
(2017), conclude that financial education interventions only seem to have a sub-
stantially small impact on the financial behavior of disadvantaged groups. As it is
challenging and not without irony to provide a survey on meta studies, we approach
this task from a different angle and mainly review methodological particularities.
Meta-studies give a summary of the direction and magnitude of parameter es-
timates for a well-defined and narrowed research question. Measurement of ”effect
sizes” is the key to meta-analysis. They measure and weigh all available and com-
parable estimations within and across studies. From systematic methodological
research on meta-analysis (see Lipsey and Wilson (2001), who published a leading
textbook) we know that there are many definitions of effect sizes such as the stan-
dardized mean difference, Cohen’s d, Cohen’s w, the Odd’s Ratio, Hedge’s g, and
the partial correlation r. Different meta-analytic studies use different definitions
such that comparisons between different meta-analysis are difficult.
This is also true for Fernandes et al. (2014) and Kaiser and Menkhoff (2017), who
systematically summarize the conclusions based on 77 and 126 studies, respectively.
They analyze the effect of financial education on financial behavior and decisions10.
Fernandes et al. (2014), using partial r as effect size indicator, report that financial
education explains only 0.1% of the variance in financial behaviors studied (deduced
from a calculated average partial r = 0.032). They consider the (statistically signifi-
cant) partial r of 0.032 as being very small, because “By social science and education
conventions, r < 0.10 is a small effect size, 0.10 < r < 0.40 is medium, and r > 0.40
is large.” Unfortunately, the authors do not quote any reference for this rule of
thumb for the partial r which, to the best knowledge of the authors of this survey,
has not been covered by known guidelines such as the ones by Cohen (1992) and
subsequent papers. The partial R2 (or partial r) is a problematic concept because it
heavily depends on the number and explanatory power of already included control
variables11. A model estimation plagued by omitted variable bias would create a
10Both studies also comprise the link between financial literacy and financial performance whichis not covered here, as research indicates that it is most likely plagued by endogeneity problems.
11Fernandes et al. (2014) mention a long list of controls including saving, planning for retirement,absence of debt, stock ownership and investment decisions, cash flow management, activity inretirement plans, and financial inertia such as choice of default options and payment of unnecessary
12
high partial improvement of the model fit, whereas well-designed studies with an
extensive list of control variables only show small “effects” after including financial
education instruments.
Different from Fernandes et al. (2014), Kaiser and Menkhoff (2017) use Hedge’s
g as effect size. As Cohen’s d, this is a standardized mean difference between edu-
cated and non-educated groups, but with some bias correction for the standardizing
standard error. Cohen (1992) considers such effect sizes as “small” when they are
smaller than 0.20, effect sizes around 0.50 indicate a “medium effect” and effect sizes
exceeding 0.80 represent “large effects12.
Not with standing the usefulness of such simple rules, already the difference be-
tween used rule-of-thumb-levels in both studies indicates that effect sizes should not
be directly compared. Kaiser and Menkhoff report the overall result that financial
education has an average effect size of g=0.086 on financial behavior, using a sample
of 349 effect sizes reported by 90 studies. Thus, according to the usual standards of
standardized differences, the impact of financial education on financial behavior can
be considered as “small”. Kaiser and Menkhoff (2017, p.9) comment their average
effect size of g=0.086 as being “more favorable” than the one by Fernandes et al.
(2014). However, as described above, effect sizes are not directly comparable such
that the reason for their conclusion remains unclear.
Summing up the main findings of both meta analysis, the overall result is that
included studies reveal a statistically significant yet substantially small impact of
financial education on financial behavior. However, the interpretation of overall
averages is limited. As likewise stated by Miller et al. (2015), the great heterogene-
ity of results makes it impossible to calculate a meaningful effect size as a general
benchmark. Both meta-studies, especially the one by Kaiser and Menkhoff (2017),
also look at different subsamples of which the subsample of low-income is of pri-
mary interest for this survey. Both studies find that for the low-income group the
size effect is even significantly below the (small) average effect. In the Kaiser and
Menkhoff paper, the effect would be reduced to g=0.02 (0.086-0.065, see Table 3
in Kaiser and Menkhoff (2017)), and Fernandes et al. (2014) report a low-income
specific partial r of 0.025.
The best way to evaluate the effects of financial education interventions is to com-
pare the difference of financial outcomes between randomly allocated “treated” and
fees.
12This rule is also quoted by Kaiser and Menkhoff (2017).
13
“control” groups (Collins and O’Rourke (2010)). According to Kaiser and Menkhoff
(2017, Table 3), the estimated effect sizes of all included RCTs is g = 0.016 (0.086 -
0.070). Results by Fernandes et al. (2014) confirm that the more rigorous random-
ized research design leads to a strong decrease in effect sizes. They report a tiny
partial r of 0.009. Not surprisingly, when focusing on RCTs for the sub-group of
low-income clients, the effect even becomes negative. Kaiser and Menkhoff (2017,
Table 3) use the evidence from 44 RCT studies and report that the effect size of
considered low-income clients is by -0.066 smaller than the overall effect found in
RCT studies.
The advantage of meta-analysis is that they give a comprehensive picture and
lead to a better understanding of the variation of existing results in the literature,
but they may not be helpful when looking for the “true” effect. Given that “most
published research findings are false” (Ioannidis (2005)), a meta-analysis runs the
risk of just reflecting and propagating common misspecifications in existing research
papers13. Koetse et al. (2013) have run Monte Carlo simulations to analyze the
effects of these misspecifications on results of a meta-analysis. When researchers
try to identify a “true” effect rather than summarizing the prevailing variation in
outcomes, Koetse et al. (2013) show the effectiveness of including dummy variables
to control for obvious primary study misspecification.
Somewhat surprisingly, except hinting at the differences between RCT and Non-
RCTs, misspecification does not seem to play a prominent role in the meta-study by
Kaiser and Menkhoff (2017). Fernandes et al. (2014) distinguish between OLS and
2SLS results and they address the problem of the omitted variable bias by conduct-
ing three additional primary data studies (but they do not cover misspecification in
the meta-analysis itself). In the discussion paper by Kaiser and Menkhoff (2017),
so far there is no mention of expressions like ”endogeneity”, ”omitted variables” or
”2SLS” which could potentially be associated with the analysis of misspecification
and its consequences.
Thus, even evidence from rich meta-studies does not free us from having a deeper
look at single papers and reported effects by target group, estimation method and
model specification.
13For instance, Bruns et al. (2014) show the misleading consequences of the common practiceof overfitting in small samples for meta-analytic studies. In a meta-analysis of the validity of thedeterrence hypothesis, Dolling et al. (2009) point out that estimations change considerably bychoice of control variables.
14
4.2 Empirical Findings by Targeted Groups
Considering the fact that financial education interventions are usually aimed at
improving the welfare of disadvantaged groups, we organize this subsection by the
target groups. More specifically, we review studies on financial education and the
immigrants/migrants, the low-income earners and the young. A study is picked only
when the intervention is explicitly targeted at the corresponding group.
We find that studies using large-scale samples are mostly on financial education
and the young, while the number of studies on financial education and immigrants is
still small but increasing. Financial education targeted at the low-income group has
been implemented more often in less-developed countries. Our list is not complete
but rather a suggestive selection of studies and only published studies are selected.
Other research under the scope but not listed includes studies on the effects of fi-
nancial education on women (Goldsmith and Goldsmith (2006), Field et al. (2010),
financial education at the workplace (Bernheim and Garrett (2003), Drexler et al.
(2014)); financial education on selected participants (Agarwal et al. (2009), Skim-
myhorn et al. (2016), etc.).
A quick look from Table 1 to Table 3 reveals that there are two main method-
ological approaches to study the effects of financial education on the disadvantaged
groups. The first one is the use of randomized controlled trials (RCT) to identify the
causal effects of an intervention by randomly assigning it to the participants. The
second one is to apply thorough econometric techniques to disentangle the causal
effects of quasi-experiments of financial education interventions. Usually, the former
method covers a smaller sample size, given the costs of implementing RCTs.
As addressing the target group of immigrants requires more effort than address-
ing school students, studies on the former group are based on relatively small sam-
ples, and they use more often RCTs than econometric techniques, which would
require larger sample sizes. All of the selected studies on financial education and
migrants and immigrants (Table 1) employ the RCT method. Many studies focus
on the effects on remittance behavior. The reason for this increasing attention to
the topic could be that remittance flows to developing countries become more sig-
nificant, as documented by studies like Mohapatra et al. (2011) and Doi et al. (2014).
Seshan and Yang (2014) and Gibson et al. (2014) find significant effects of fi-
nancial education neither on remittance behavior nor on savings, and also Barcellos
et al. (2016) only report a short-lived impact fading away after about six months.
15
Results by Doi et al. (2014) suggest that it might matter whether the program makes
use of family peer effects or not. Their interesting finding is that training both the
migrant and the family member of the worker who would be responsible for receiv-
ing remittances together has large and significant impacts on knowledge, behavior,
and savings. Training the migrant only has very little impact on remaining family
members.
To improve the welfare of low-income earners is often on the agenda of policy
makers. Also (central) banks care about the economic status of the low-income
group to avoid defaults on credit files and to stabilize the economy. Accordingly,
policy makers (e.g., Grinstein-Weiss et al. (2015)) and financial institutions (see,
e.g., Hartarska and Gonzalez-Vega (2005), Agarwal et al. (2009)) are motivated to
implement intensive financial educational programs targeted at this disadvantaged
group. In those studies, financial education counselling programs are often provided
to people who are identified to be in rather poor economic conditions, and econo-
metric techniques such as the propensity score matching are applied to estimate the
effects. Almost all of the studies in Table 2, except the one by Collins (2013), find
that financial education interventions have positively significant effects on the low-
income earner’s saving and default rates. Collins (2013) detects significant effects
for self-reported behavior, which, however, do not correspond with actual behavior
as savings or credit behavior do not change substantially.
Table 3 summarizes some studies on financial education and the young. Most
studies find significant, even long-term, effects of financial education (e.g., Batty
et al. (2015) and Brown et al. (2016)). There are more studies on this topic but
not included for reasons of brevity. More strictly speaking, the young should not
be called “disadvantaged” per se, as “youth” is a transient period of everyone’s life.
None of the studies included in our Tables explicitly focuses on the more disadvan-
taged subgroup of young people.
However, as emphasized by Atkinson and Messy (2015) and OECD (2017), the
young are also one of the at-risk segments of the population, and attitudes and
routines developed during youth and adolescence seem to strongly influence adult
financial well-being or disadvantage. Also, theoretical contributions based on life
cycle models (see above) suggest that financial education on the young could improve
the social welfare of members of disadvantaged families over a lifetime and, thus,
might reduce economic inequality. Another reason why we care about financial
education and the young is that large-scale financial education interventions are
accessible to policy makers and appealing to scholars, although the discussion on
16
the upsides and downsides of those interventions is still open.
17
Tab
le1:
Fin
anci
alE
duca
tion
and
the
Imm
igra
nts
/Mig
rants
Au
thor
Yea
r o
f
Inte
rven
tio
n
Inte
rven
tion
/Data
set
Sam
ple
M
eth
od
F
ind
ings
Ses
han
and
Yan
g (
20
14
) 20
10
Sin
gle
ses
sio
n o
f a
few
hou
rs;
foll
ow
-up s
urv
ey
over
a y
ear
late
r
15
7 i
n t
he
trea
tmen
t
gro
up
, 7
5 i
n t
he
con
trol
gro
up;
India
n m
arri
ed
mal
e m
igra
nts
work
ing
in Q
atar
RC
T
Wiv
es o
f tr
eate
d m
igra
nts
bec
ame
mo
re l
ikel
y t
o s
eek o
ut
finan
cial
edu
cati
on t
hem
selv
es.
Tre
ated
mig
rants
an
d t
hei
r w
ives
bec
ame
more
lik
ely t
o
mak
e jo
int
dec
isio
ns
on m
oney
mat
ters
. N
o s
ign
ific
ant
impac
ts o
n s
avin
gs
or
rem
itta
nce
s, h
igh
tre
atm
ent
effe
ct
het
ero
gen
eity
.
Doi
et a
l.
(201
4)
20
10
18
ho
urs
over
tw
o d
ays
for
mig
rants
, 8
hours
ov
er t
wo d
ays
for
mig
rant
fam
ily m
ember
s;
3 r
ounds
of
foll
ow
-up
surv
eys
(20
11.0
3,
2011.0
9,
20
12.0
1)
40
0 m
igra
nt
hou
seh
old
s
in I
ndon
esia
R
CT
Tra
inin
g b
oth
the
mig
ran
t an
d t
he
fam
ily m
ember
Res
pon
sible
for
rece
ivin
g r
emit
tan
ces
toget
her
has
lar
ge
and
sig
nif
ican
t im
pac
ts o
n k
now
ledge,
beh
avio
r, a
nd
savin
gs.
Tra
inin
g o
nly
th
e m
igra
nt
has
no i
mp
act
on t
he
rem
ainin
g
fam
ily m
ember
s.
Th
e te
ach
able
tim
e m
atte
rs f
or
the
mag
nit
ud
e of
the
effe
ct.
Gib
son e
t al
.
(201
4)
20
11
2 h
our
single
ses
sion;
3
mo
nth
ly f
oll
ow
-up
surv
eys,
on
e fi
nal
rou
nd
6 m
on
ths
late
r af
ter
the
inte
rven
tion
349
Pac
ific
isla
nder
s, 3
52
east
Asi
ans,
209
Sri
Lan
kan
s in
Aust
rali
a/N
ew Z
eala
nd
RC
T
Th
e tr
ainin
g i
ncr
ease
s fi
nan
cial
kno
wle
dge
and
info
rmat
ion
-see
kin
g b
ehav
ior
and r
educe
s th
e ri
sk o
f
swit
chin
g t
o c
ost
lier
rem
itta
nce
pro
du
cts.
Tra
inin
g d
oes
not
resu
lt i
n s
ignif
ican
t ch
anges
in t
he
freq
uen
cy a
nd a
moun
t of
rem
itta
nce
s.
Bar
cell
os
et a
l.
(201
6)
20
12
A
sin
gle
ses
sio
n o
f
onli
ne
trai
nin
g
563
fir
st a
nd s
econd
-
gen
erat
ion i
mm
igra
nts
to
the
US
A a
nd
2,4
98
nat
ives
RC
T
Tre
ated
im
mig
rants
an
d t
hei
r ch
ildre
n w
ere
mo
re l
ikel
y t
o
answ
er f
inan
cial
kno
wle
dge
qu
esti
ons
corr
ectl
y r
ight
afte
r
the
inte
rven
tion
. T
he
effe
cts
faded
aw
ay s
ix m
onth
s la
ter.
No s
ign
ific
ant
effe
ct o
n t
he
trea
ted i
mm
igra
nts
’ fi
nan
cial
beh
avio
r(s)
.
18
Tab
le2:
Fin
anci
alE
duca
tion
and
the
Low
-inco
me
Ear
ner
s
Au
tho
r Y
ear
of
Inte
rven
tio
n
Inte
rven
tion
/Da
tase
t
Sam
ple
Siz
e M
eth
od
F
ind
ings
Har
tars
ka
and
Veg
a
(200
5)
199
2 -
20
00
Counse
llin
g p
rogra
m
imple
men
ted
in s
ever
al m
id-
wes
t st
ates
in
th
e U
S
(Co
mm
un
ity M
ort
gag
e L
oan
Pro
gra
m)
392
ob
serv
atio
ns,
wit
h 2
78
co
un
sele
d
and
11
4
no
nco
unse
led
loan
s
Co
x
pro
port
ion
al
haz
ard
mod
el
Het
ero
gen
eous
find
ings:
Fo
r
rura
l an
d s
ubu
rban
and f
or
non
-met
ro,
low
-
inco
me
ho
use
hold
s, t
his
cou
nse
ling p
rogra
m
do
es n
ot
signif
ican
tly d
ecre
ase
def
ault
rat
es.
It m
atte
rs i
n u
rban
, in
ner
cit
ies
envir
onm
ents
.
Ag
arw
al e
t
al.
(201
0)
200
5 -
20
07
Pro
gra
m (
INH
P)
des
ign
ed t
o
assi
st l
ow
- an
d m
od
erat
e-in
com
e
ho
use
hold
s in
purs
uit
of
hom
e
ow
ner
ship
. T
hre
e-h
our
clas
s o
n
mo
ney
man
agem
ent
pra
ctic
es;
mo
nth
ly o
ne-
on
-on
e m
eeti
ngs
up
to t
wo
yea
rs
359
wer
e tr
eate
d i
n
the
mu
ch l
arger
who
le s
ample
(dat
a
pro
vid
ed b
y I
NH
P)
Pro
pen
sity
score
mat
chin
g
Low
-and
-mo
der
ate
finan
cial
edu
cati
on p
rogra
m
gra
duat
es s
how
ed s
ubst
anti
ally
low
er
del
inqu
ency r
ates
on h
om
e m
ort
gag
es.
Co
llin
s
(20
13)
20
12
Man
dat
ory
fin
anci
al e
du
cati
on
pro
gra
m f
or
low
-in
com
e
fam
ilie
s in
a s
ub
sidiz
ed
housi
ng p
rogra
m;
self
-
report
ed s
urv
ey 1
2 m
on
ths
late
r
14
4
pro
pen
sity
sco
re
mat
chin
g;
two
-sta
ge
esti
mat
es
Imp
rov
emen
t on
ly i
n s
elf-
report
ed b
ehav
ior
of
the
low
-inco
me
fam
ilie
s, b
ut
no
mea
sura
ble
eff
ects
on t
hei
r sa
vin
gs
or
cred
it.
Gri
nst
ein
-
Wei
ss e
t
al. (2
015
)
199
7 -
20
01
Ev
aluat
ion o
f In
div
idu
al
Dev
elop
men
t A
ccou
nt
(ID
A,
targ
eted
at
the
poo
r). S
elf-
sele
cted
and
pro
gra
m-s
elec
ted
par
tici
pan
ts o
f A
mer
ican
Dre
am
Poli
cy D
emonst
rati
on
(A
DD
).
Mea
n a
tten
dan
ce i
s ab
out
10
ho
urs
.
2,0
44
Pro
pen
sity
sco
re
mat
chin
g
Rel
ativ
e to
cou
nte
rpar
ts w
ho d
id n
ot
com
ple
te
edu
cati
on
al r
equir
emen
ts,
IDA
par
tici
pan
ts (
in
par
ticu
lar
those
aged
36
and o
lder
) w
ho
com
ple
ted p
rogra
m r
equir
emen
ts f
or
fin
anci
al
educa
tio
n h
ad h
igher
sav
ings.
19
Tab
le3:
Fin
anci
alE
duca
tion
and
the
You
ng
Au
tho
r Y
ear
of
Inte
rven
tion
In
terv
enti
on
/Data
set
Sam
ple
M
eth
od
F
ind
ings
Bat
ty e
t
al.
(20
15
)
201
1-2
01
2,
20
12
-20
13
Fiv
e w
eekly
les
son
s of
appro
xim
atel
y 4
5 m
inute
s ea
ch,
focu
sed o
n s
avin
gs,
fin
anci
al
dec
isio
n m
akin
g, m
oney
man
agem
ent
for
gra
des
4 t
hro
ugh
5 i
n W
isco
nsi
n,
US
; fo
llow
-up
surv
eys
App
rox
imat
ely
1,5
00
stud
ents
in t
he
firs
t per
iod,
760
stud
ents
in t
he
seco
nd p
erio
d
RC
T
Even
one
yea
r la
ter,
stu
den
ts
expo
sed t
o f
inan
cial
educa
tion h
ave
more
posi
tive
atti
tudes
tow
ard
s
per
sonal
fin
ance
and a
pp
ear
mo
re
lik
ely t
o s
ave.
Ber
nh
eim
et
al.
(200
1)
95
7-1
985
L
egis
lati
on m
and
atin
g c
onsu
mer
edu
cati
on i
n s
econd
ary s
chools
in
29 s
tate
s of
the
US
2,0
00
resp
ond
ents
aged
30
-
49 i
n 1
999
Pro
bit
/OL
S r
egre
ssio
n
bas
ed o
n p
op
ula
tion
ran
ks
of
dep
end
ent
var
iable
s, m
edia
n
regre
ssio
n,
Man
dat
ed h
igh s
choo
l cu
rric
ulu
m h
as
rais
ed b
oth
ex
po
sure
to f
inan
cial
curr
icula
and s
ubse
qu
ent
asse
t
accu
mula
tion
on
ce e
xpose
d s
tuden
ts
reac
hed
adult
ho
od.
Bro
wn
et
al.
(20
16)
19
00s-
20
00
s
Var
iati
on i
n t
he
tim
ing o
f th
e
enac
tmen
t of
fin
anci
al e
duca
tio
n
and m
ath
emat
ics
refo
rms
in h
igh
sch
oo
l cu
rric
ula
acr
oss
and
wit
hin
US
sta
tes,
Co
unci
l fo
r E
conom
ic E
du
cati
on
surv
ey,
Key S
tate
Edu
cati
on
Poli
cies
on
PK
-12 E
duca
tio
n
surv
ey,
FR
BN
Y C
onsu
mer
Cre
dit
Pan
el
Aro
un
d 7
mil
lio
n
obse
rvat
ion
s
DID
est
imat
ion a
nd
even
t-st
ud
y c
ontr
oll
ing
stat
e sp
ecif
ic a
nd
coho
rt-s
pec
ific
tim
e
tren
ds
Mat
h a
nd
fin
anci
al l
iter
acy e
duca
tion
mo
des
tly r
edu
ce t
he
inci
den
ce o
f
adver
se o
utc
om
es;
econo
mic
educa
tion
lead
s to
a d
ecli
ne
in y
outh
s' a
ver
age
risk
sco
res.
20
Tab
le3:
Fin
anci
alE
duca
tion
and
the
You
ng
(Con
tin
ued
)
Au
tho
r Y
ear
of
Inte
rven
tion
In
terv
enti
on
/Data
set
Sam
ple
M
eth
od
F
ind
ings
Bru
hn
et
al.
(20
16)
20
10
-20
11
Hig
h s
cho
ol
fin
anci
al e
du
cati
on
pro
gra
m f
or
thre
e se
mes
ters
span
nin
g 6
sta
tes,
892 s
cho
ols
in
Bra
zil,
foll
ow
-up s
urv
eys
44
0 t
reat
men
t
and 4
52 c
ontr
ol
scho
ols
RC
T
The
pro
gra
m i
ncr
ease
d s
tuden
t
finan
cial
pro
fici
ency b
y a
quar
ter
of
a st
and
ard
dev
iati
on a
nd
rai
sed
gra
de-
level
pas
sin
g r
ates
; m
ixed
effe
cts
on s
hort
-ter
m f
inan
cial
beh
avio
r. S
ignif
ican
t
inte
rgen
erat
ional
eff
ects
of
the
pro
gra
m o
n p
aren
ts
Cole
et
al.
(20
16)
19
00s-
20
00
s
Sta
te m
and
ated
fin
anci
al e
du
cati
on
,
and
mat
hem
atic
s re
form
s in
the
US
,
20
00 U
.S.
Cen
sus,
po
ole
d p
anel
s of
FIP
P, F
RB
NY
Co
nsu
mer
Cre
dit
Pan
el
Aro
un
d 6
mil
lio
n
ob
serv
atio
ns
DID
est
imat
ion a
nd
even
t-st
ud
y c
ontr
oll
ing
for
stat
e fi
xed
eff
ects
Ad
dit
ional
mat
hem
atic
al t
rain
ing
inst
ead o
f fi
nan
cial
educa
tion,
lead
s to
gre
ater
fin
anci
al m
arket
par
tici
pat
ion
,
inves
tmen
t in
com
e, a
nd b
ette
r cr
edit
man
agem
ent.
Lü
hrm
ann
et a
l. (
20
15)
201
2
Thre
e 90
min
ute
s m
od
ule
s fo
cusi
ng
on
sh
oppin
g, pla
nn
ing,
sav
ing,
and
targ
eted
at
scho
ol
stu
den
ts a
ged
mai
nly
bet
wee
n 1
3 a
nd 1
5 y
ears
old
in G
erm
any;
surv
eys
bef
ore
and
afte
r tr
ain
ing
52
1 s
tud
ents
in
trea
tmen
t an
d 1
15
in
con
trol
for
an a
fter
-
trai
nin
g p
erio
d
RC
T
Tra
inin
g p
rogra
m s
ignif
ican
tly
incr
ease
s te
enag
ers’
in
tere
st i
n
finan
cial
mat
ters
and t
hei
r fi
nan
cial
kn
ow
ledge,
esp
ecia
lly t
hei
r ab
ilit
y t
o
pro
per
ly a
sses
s th
e ri
skin
ess
of
asse
ts.
21
4.3 Common Issues in RCTs
It is perhaps not surprising to find even less significant effects of financial education
on immigrants/migrants than on low-income earners, as reported RCT interventions
for migrants and immigrants summarized in Table 1 are shorter and less intensive
than those for low-income groups summarized in Table 2. To avoid unobserved bias
caused by more intensive and lasting interventions, such as a decreasing participa-
tion rate, changing characteristics of participants (Glewwe (2002)) and too much
variety of teaching materials, it would indeed be favorable to apply shorter sessions
of financial education through more controlled methods. The downside of the purity
of RCTs is that the education interventions were possibly too weak and short such
that they kept below a critical threshold of efficacy.
Different from the smaller samples in RCTs on financial education and immi-
grants/migrants, RCTs applied to the young, especially to school students, can be
much easier applied to larger samples and based on more intensive treatment. Re-
markable examples are Berry et al. (2015) for Ghana, Bjorvatn et al. (2015) for
Tanzania, Luhrmann et al. (2015) for Germany, Batty et al. (2015) for the U.S., and
Becchetti and Pisani (2011) and Becchetti et al. (2013) for Italy. Likewise in con-
trast to RCT results found for immigrants/migrants, many studies report positive
significant effects, for instance on students’ attitudes regarding personal financial
and saving behavior (Batty et al. (2015)), financial proficiency (Bruhn et al. (2016))
or financial knowledge and risk aversion (Luhrmann et al. (2015)).
As also pointed out by Glewwe (2002), RCTs can be the ideal way to identify the
causal effects of financial education on economic behaviors, given proper random-
ization. He highlights the sample selection issue as potential caveat. Bruhn et al.
(2013) indicate that one common issue faced by all small-scale RCTs is that the
selection of participants is usually based on a voluntary process, and the voluntary
participation is typically low.
For example, the RCT on migrants from Indonesia in Doi et al. (2014) seems to
be faced with the sample selection issue. The randomized selection process started
with a diagnostic study to identify a region with a high concentration of migrants
(East Java), the sample was organized by Malang’s Manpower and Transmigration
Office, and the participants have been selected by a privately-owned Indonesian
Manpower Placement Company. The process ended up with almost all participants
being female which reflects the common existence of non-unitary households with
women deciding on daily spending. A further issue is that no information was given
22
on the response ratio of people who agreed to participate to people who were invited.
In addition, those who participated were selected from those who were still staying
in the dormitory facilities of the recruitment agencies or lived close by. Thus, it
seems rather problematic to extract external validity from the study – but this a
rather general problem with many other RCT studies and not limited to Doi et al.
(2014), as emphasized by Cartwright (2007) and Deaton and Cartwright (in press).
RCT, too, is no gold standard.
Gibson et al. (2014) spend a lot of effort on designing the content of the in-
tervention in Australia, and on assuring that the educational content is sufficiently
accessed by participants14. Follow-up surveys were conducted for multiple times,
which allows them to examine the dynamic effects of the educational intervention15.
The intervention was carefully designed in many aspects. However, the participants
were not selected in a randomized way but rather through social networks. For
example, one third of Pacific Island participants were recruited from a Pacific cul-
tural festival, and the rest were from the main Pacific outdoor market or churches.
Besides that, all participants were drawn from Hamilton and Auckland, and no ev-
idence indicates that this is a randomized draw for a well-defined population. On
the positive side, the experiment represents an innovative approach for how to im-
plement a financial education intervention for migrants with low costs.
A similar ambiguity exists in Barcellos et al. (2016) on financial literacy targeted
at immigrants in the US, even though much effort has been invested to assure the
randomization of the sample as much as possible, and to enable reasonable inter-
pretations of results. However, as also mentioned by the authors themselves, the
interpretation is limited by the small sample size (135 in treatment group 1, 118
in treatment group 2 and 117 in the control group, only 330 in the follow-up sur-
vey), and by attrition problems. Besides that, the participants are recruited from
an Internet panel of respondents 18 years and older who agreed to participate in
occasional online surveys. This self-selection bears the risk of receiving responses
from a particular (unknown) subgroup of the population, once again questioning
the external validity of results. Moreover, there were two sets of financial educa-
tional materials targeted at immigrants/migrants which were released to treatment
groups online, but no clear description on the heterogeneity of the training intensity
was given, i.e., how much time participants spent on the material for their training
14Written materials were provided to trainees; one training session was two hours and usuallyheld at churches, community centers, and sports clubs for 30 people at one time.
15Monthly follow-up surveys were conducted for the three months after the training interventionto capture the short-time effects of the training intervention. One more follow-up survey was donein the sixth month after the training, to measure its long-term effects.
23
sessions. A particular problem seems to be the large variety of English language
skills. Those with poor skills underperform because the survey questions were put
in English such that any test of financial knowledge is at the same time an English
proficiency test and measures language gaps. This joint hypotheses test makes it
difficult to identify the ”true” effect of financial education because test results are
blurred by the heterogeneity of language skills.
Looking at Glewwe (2002)’s list of potential problems with RCTs, the two issues
of non-stable characteristics of sampled individuals and of sample selection biases are
still relevant. Even though Bruhn et al. (2016) use one of the largest samples among
those targeted at the young, the RCT is still faced with both problems. In regard to
the first issue, there is not much information about whether parents of the “treated”
students decrease their efforts to support the students during the training period or
not. In terms of the second issue, the description of the study shows that the RCT is
well designed, but there is still much ambiguity about the implementation process of
the program. The RCT was implemented in public high schools in Brazil. 892 public
high schools and approximately 25,000 students from the Federal District, and 5 out
of 26 states are involved. The Secretariat of Education in each state assembled a
list of schools willing to participate in the program. Several steps were applied to
assure the randomization of the assignment16. However, we do not have information
about whether there were parents who enrolled their children into the treatment
group in the course of the intervention. Moreover, it is mentioned that also about
17 % of schools in the control group have implemented some own financial education,
but the authors do not know how intensive this training might have been. They
presume that these programs might have been less intensive than the one studied in
this paper.
4.4 Common Econometric Issues
When it comes to the econometric approach, Glewwe (2002) summarizes five com-
mon issues which could arise in this method. Though the topic of Glewwe (2002)
is on general education programs in developing countries, addressed problems also
hold for financial education programs: (1) considering the unobserved components
16The randomization of assigning schools to treatment groups and control groups was doneby the research team through a computer, but small adjustments were done manually due toadministration requirements or other reasons. A stratification method was implemented to makethe sample representative of public schools in terms of the quality of the school. Besides, GDP andsavings volume per capita of the municipality, school location, number of students and teachersin the school, school drop-out rate, school graduation rate are balanced across the treatment andcontrol groups.
24
of the participants, such as factors on motivation, inner ability; (2) omitted vari-
ables, for example, the instructors’ quality; (3) sample selection, when people with
disadvantaged family background might have fewer chances to attend schools offer-
ing curricula with mandated financial education; (4) endogeneity of participating in
a financial education intervention, for example if people with lower wealth might be
less motivated to attend the intervention; (5) measurement error in the regressors
and improper specification of the dependent variable.
A good example to address the importance of econometric issues is shown by the
comparison of two studies on financial education and the young, i.e., Bernheim et al.
(2001) and Cole et al. (2016). Both studies investigate the effects of the same fi-
nancial educational intervention but arrive at completely different conclusions. The
financial education intervention under investigation by Bernheim et al. (2001) is
a national mandatory ”consumer” education intervention in secondary schools in
the U.S. 29 states adopted it during the time period 1957 to 1985. Bernheim et al.
(2001) use a cross-sectional household survey conducted in fall 1995 to identify long-
term effects of the mandatory intervention on households’ saving behavior. Careful
empirical modeling, including a difference-in-difference approach, and robustness
checks have been conducted to assure the reliability of the model. However, the
significantly positive effects found in Bernheim et al. (2001) are “washed out” by
simply adding state fixed effects in Cole et al. (2016). Cole et al. (2016) admit
that the difference might come from the difference of data sources, but they provide
econometric evidence showing that this is not the reason. But Cole et al. (2016) are
not perfect either. No obviously convincing evidence has shown that the systematic
differences of effects between states (leading to the significance of state fixed effects
and usually attributed to unknown third factors of some sort) do not reveal the
heterogeneity of efficacy of state-specific financial education mandates (such that
the state-fixed effect would “throw out the baby with the bath water”).
Despite the critical assessment in later research, the paper by Bernheim et al.
(2001) can be considered as seminal. It had a strong impact on future work on the
effects of financial education on financial behavior such as savings. It also presents
one of the first studies examining the causal effects of financial education on cer-
tain economic behaviors. Cole et al. (2016) leave us even more skeptical about the
reliability and validity of results, not only with respect to Bernheim et al. (2001)
but also in more general terms. The comparison of the two papers reveals that the
choice of methodology matters a lot. In order to get a less biased estimation, authors
need to find the most suitable technique, adapted to the institutional context and
the nature of the data. There is not one silver bullet which could be used for all
25
potential research questions on the effect of financial education.
In sum, we need to conclude that there is no satisfying answer to the question
of what is the best way of testing the impact of financial education, RCT or the
econometric approach. RCT has become the more popular approach in recent years,
but tides might be turning. As pointed out by Deaton and Cartwright (in press):
“At best, RCT yields an unbiased estimate, but this property is of limited practical
value. Even then, estimates apply only to the sample selected for the trial, often no
more than a convenience sample, and justification is required to extend the results to
other groups”. RCT is a fine technique to study the causal effects of financial educa-
tion, and it is very popular when the targeted groups are the immigrants/migrants
and the young, but results depend on the pre-selected and reachable sub-population
of the randomization process and the treatment. Financial education interventions
are way far from being consistent and comparable with each other.
On the other hand, empirical results from using non-RCTs are also mixed, and
not necessarily superior to the RCT approach. But there is evidence that the applied
econometric approach is getting more mature. For example, in addition to common
controls like age and gender, other relevant elements like personal motivation (Man-
dell and Klein (2009)) or time preferences (Meier and Sprenger (2015)) are taken
into consideration. On the methodological side, there is new literature (reviewed by
Belloni et al. (2017) and Athey and Imbens (2017) among others) proposing machine
learning techniques or high-dimensional settings, which aim to develop estimation
and inference tools in program evaluation allowing for a large number of control
variables (much larger than the sample size). However, at the present stage, it is
too early to say anything about the usefulness in practical applications.
The lesson to be learned is that diverse research methods should be combined in
a reasonable way such that evaluation studies can play a role in building scientific
knowledge. RCTs have the potential to estimate clean average treatment effects,
but in order to be comparable with each other they need to adjust for particularities
in institutional settings and include prior knowledge. Investigations also need to
build on theoretical work, and conclusions should go beyond simple statements on
“what works”, but “why things work” (Deaton and Cartwright (in press)).
26
5 Concluding Remarks
Financial policy worldwide welcomes and embraces initiatives to improve finan-
cial literacy. Financial education programs are considered as suitable measures to
develop the knowledge, skills, and ability to make safe financial decisions and to
increase resilience against future financial shocks. As many scholars and politicians
see social inequality as the source of the 2008 financial crisis (known as the Rajan
hypothesis, inspired by Rajan (2011)), and because financial education and financial
literacy are seen as the key to solve the problem of unequal chances, our survey has
a focus on the group at the bottom of the financial wealth distribution. However,
in sharp contrast to the popularity of financial education interventions, studies on
the effectiveness of interventions show mixed results for the disadvantaged groups
under consideration, and particularly for the group of migrants almost no sustain-
able impact on financial behavior has been detected and reported by the literature.
Our survey reviews both the theoretical and empirical findings in order to un-
derstand why this discrepancy exists. The survey first highlights the problematic
link between financial education, financial literacy and financial behavior in the
existing literature, and discusses the importance of third factors such as cognitive
ability, mathematical capability or numeracy. Theories on financial literacy help us
to understand the heterogeneous effect of interventions across the population and
over the life cycle which in turn could lead to better identification, targeting and
implementation of pinpointed financial education interventions.
Looking at empirical results, we find that the young, the immigrants/migrants
and the low-income groups obtain more attention from scholars than other disad-
vantaged groups. Studies on large-scale samples, both using RCTs and econometric
methods for quasi-experiments, show up more often for investigating financial edu-
cation on the young. The effect on remittance behavior is the focus in studies on
samples of migrants. Some support in favor of financial education for the young has
been detected in terms of higher short-term financial knowledge and awareness in
some studies, but there is no proven evidence of improved long-term behavior after
growing up. From the methodological point of view, despite the criticism brought
forward by Deaton and Cartwright (in press) and others, RCT has become a popular
approach to evaluate the effect of financial education in recent years, in particular
for studies on remittance behavior of migrants. Results based on RCT reveal even
smaller effects than those based on the econometric approach. Our survey discusses
some limitations of both RCTs and econometric studies which might be taken into
account by future research. We conclude that further standardizations of RCTs
27
would enhance the information transparency and comparability of results from eval-
uation studies. Moreover, we give a critical assessment of meta-analytic insights
regarding financial education which bear the risk of just reflecting and propagating
common misspecifications in existing research papers.
In sum, despite the enthusiasm of politicians, who see financial education and
financial literacy as the key to tackle the problem of financial vulnerability, so far,
there is no clear evidence, or at least no scientific consensus, on the effectiveness
of interventions. The financial education fallacy addressed by Willis (2011) alerts
us that the constant search for effective financial education might cost enormously
such that the costs of financial education programs might outweigh potential ben-
efits. Even if people promoting financial education don’t agree with Willis (2011),
they should have second thoughts before spending scarce governmental resources.
It is perhaps true that the financial behavior of the poor is arguably more con-
trolled by lack of aspirations such that financial policy should be directed towards
behaviorally motivated anti-poverty policies (as described by Bertrand et al. (2004)).
Given our approach to the literature, a limitation of the survey is that we cannot
review all the studies of interest in detail. Besides, we do not examine the inter-
ventions by countries, although country and state-specific heterogeneity might also
be important to understand the effectiveness of financial education programs for
disadvantaged groups.
28
References
Agarwal, S., P. M. Skiba, and J. Tobacman (2009). Payday loans and credit cards:
New liquidity and credit scoring puzzles? The American Economic Review 99 (2),
412–417.
Arrow, K. J. (1987). The demand for information and the distribution of income.
Probability in the Engineering and Informational Sciences 1 (1), 3–13.
Athey, S. and G. W. Imbens (2017). The state of applied econometrics: Causality
and policy evaluation. Journal of Economic Perspectives 31 (2), 3–32.
Atkinson, A. and F. Messy (2015). Financial education for migrants and their
families. Technical report, OECD Publishing.
Attanasio, O. P. (1999). Consumption. Handbook of macroeconomics 1, 741–812.
Barcellos, S. H., L. S. Carvalho, J. P. Smith, and J. Yoong (2016). Financial educa-
tion interventions targeting immigrants and children of immigrants: Results from
a randomized control trial. Journal of Consumer Affairs 50 (2), 263–285.
Batty, M., J. M. Collins, and E. Odders-White (2015). Experimental evidence on the
effects of financial education on elementary school students’ knowledge, behavior,
and attitudes. Journal of Consumer Affairs 49 (1), 69–96.
Becchetti, L., S. Caiazza, and D. Coviello (2013). Financial education and invest-
ment attitudes in high schools: Evidence from a randomized experiment. Applied
Financial Economics 23 (10), 817–836.
Becchetti, L. and F. Pisani (2011). Financial education on secondary school students:
The randomized experiment revisited. Econometica Papers wp34, Econometica.
Belloni, A., V. Chernozhukov, I. Fernandez-Val, and C. Hansen (2017). Program
evaluation and causal inference with high-dimensional data. Econometrica 85 (1),
233–298.
Ben-Porath, Y. (1967). The production of human capital and the life cycle of earn-
ings. Journal of Political Economy 75 (4, Part 1), 352–365.
Bernheim, B., D. M. Garrett, and D. M. Maki (2001). Education and saving: The
long-term effects of high school financial curriculum mandates. Journal of Public
Economics 80 (3), 435 – 465.
Bernheim, B. D. and D. M. Garrett (2003). The effects of financial education in
the workplace: Evidence from a survey of households. Journal of Public Eco-
nomics 87 (7), 1487–1519.
29
Berry, J., D. Karlan, and M. Pradhan (2015). The impact of financial education for
youth in Ghana. WorkingPaper 15-043/V, Tinbergen Institute.
Bertrand, M., S. Mullainathan, and E. Shafir (2004). A behavioral-economics view
of poverty. American Economic Review 94 (2), 419–423.
Bjorvatn, K., A. W. Cappelen, L. H. Sekei, E. Sorensen, and B. Tungodden (2015).
Teaching through television: Experimental evidence on entrepreneurship educa-
tion in Tanzania.
Brown, M., J. Grigsby, W. van der Klaauw, J. Wen, and B. Zafar (2016). Financial
education and the debt behavior of the young. The Review of Financial Stud-
ies 29 (9), 2490–2522.
Bruhn, M., L. de Souza Leao, A. Legovini, R. Marchetti, and B. Zia (2016). The
impact of high school financial education: Evidence from a large-scale evaluation
in Brazil. American Economic Journal: Applied Economics 8 (4), 256–295.
Bruhn, M., G. L. Ibarra, and D. McKenzie (2013). Why is voluntary financial
education so unpopular? Experimental evidence from Mexico. Technical report,
The World Bank.
Bruns, S. B., C. Gross, and D. I. Stern (2014). Is there really granger causality
between energy use and output? The Energy Journal 35 (4), 101–133.
Campbell, J. Y. (2006). Household finance. The Journal of Finance 61 (4), 1553–
1604.
Cartwright, N. (2007). Are RCTs the gold standard? BioSocieties 2 (1), 11–20.
Cohen, A. (1992). Antecedents of organizational commitment across occupational
groups: A meta-analysis. Journal of Organizational Behavior 13 (6), 539–558.
Cole, S., A. Paulson, and G. K. Shastry (2014). Smart money? The effect of
education on financial outcomes. The Review of Financial Studies 27 (7), 2022–
2051.
Cole, S., A. Paulson, and G. K. Shastry (2016). High school curriculum and financial
outcomes: The impact of mandated personal finance and mathematics courses.
Journal of Human Resources 51 (3), 656–698.
Collins, J. M. (2013). The impacts of mandatory financial education: Evidence
from a randomized field study. Journal of Economic Behavior & Organization 95,
146–158.
30
Collins, J. M. and C. M. O’Rourke (2010). Financial education and counseling—still
holding promise. Journal of Consumer Affairs 44 (3), 483–498.
Cunha, F. and J. Heckman (2007). The technology of skill formation. American
Economic Review 97 (2), 31–47.
Deaton, A. and N. Cartwright (in press). Understanding and misunderstanding
randomized controlled trials. Social Science & Medicine.
Delavande, A., S. Rohwedder, and R. J. Willis (2008). Preparation for retirement,
financial literacy and cognitive resources. Technical report, University of Michigan
Retirement Reesearch Center Working Paper 2008-190.
Disney, R. and J. Gathergood (2013). Financial literacy and consumer credit port-
folios. Journal of Banking & Finance 37 (7), 2246–2254.
Doi, Y., D. McKenzie, and B. Zia (2014). Who you train matters: Identifying
combined effects of financial education on migrant households. Journal of Devel-
opment Economics 109, 39–55.
Dolling, D., H. Entorf, D. Hermann, and T. Rupp (2009). Is deterrence effective?
Results of a meta-analysis of punishment. European Journal on Criminal Policy
and Research 15 (1-2), 201–224.
Drexler, A., G. Fischer, and A. Schoar (2014). Keeping it simple: Financial literacy
and rules of thumb. American Economic Journal: Applied Economics 6 (2), 1–31.
Fernandes, D., J. G. Lynch Jr, and R. G. Netemeyer (2014). Financial literacy, finan-
cial education, and downstream financial behaviors. Management Science 60 (8),
1861–1883.
Field, E., S. Jayachandran, and R. Pande (2010). Do traditional institutions con-
strain female entrepreneurship? A field experiment on business training in india.
American Economic Review 100 (2), 125–29.
Gathergood, J. (2012). Self-control, financial literacy and consumer over-
indebtedness. Journal of Economic Psychology 33 (3), 590–602.
Gibson, J., D. McKenzie, and B. Zia (2014). The impact of financial literacy training
for migrants. The World Bank Economic Review 28 (1), 130–161.
Glewwe, P. (2002). Schools and skills in developing countries: Education policies
and socioeconomic outcomes. Journal of Economic Literature 40 (2), 436–482.
31
Goldsmith, R. E. and E. B. Goldsmith (2006). The effects of investment education
on gender differences in financial knowledge. Journal of Personal Finance 5 (2),
55.
Gramat,ki, I. (2017). A comparison of financial literacy between native and immi-
grant school students. Education Economics 25 (3), 304–322.
Grinstein-Weiss, M., S. Guo, V. Reinertson, and B. Russell (2015). Financial ed-
ucation and savings outcomes for low-income ida participants: Does age make a
difference? Journal of Consumer Affairs 49 (1), 156–185.
Hackethal, A., M. Haliassos, and T. Jappelli (2012). Financial advisors: A case of
babysitters? Journal of Banking & Finance 36 (2), 509–524.
Hanushek, E. A. and L. Woessmann (2008). The role of cognitive skills in economic
development. Journal of Economic Literature 46 (3), 607–68.
Hartarska, V. and C. Gonzalez-Vega (2005). Credit counseling and mortgage ter-
mination by low-income households. The Journal of Real Estate Finance and
Economics 30 (3), 227–243.
Hastings, J. S., B. C. Madrian, and W. L. Skimmyhorn (2013). Financial literacy,
financial education, and economic outcomes. Annual Review of Economics 5 (1),
347–373.
Hilgert, M. A., J. M. Hogarth, and S. G. Beverly (2003). Household financial man-
agement: The connection between knowledge and behavior. Fed. Res. Bull. 89,
309.
Hubbard, R. G., J. Skinner, and S. P. Zeldes (1995). Precautionary saving and social
insurance. Journal of Political Economy 103 (2), 360–399.
Hung, A., A. M. Parker, and J. Yoong (2009). Defining and measuring financial
literacy. Technical report, RAND CorporationWorking Papers - 708.
Huston, S. J. (2010). Measuring financial literacy. Journal of Consumer Af-
fairs 44 (2), 296–316.
Ioannidis, J. P. (2005). Why most published research findings are false. PLoS
medicine 2 (8), e124.
Jappelli, T. and M. Padula (2013). Investment in financial literacy and saving
decisions. Journal of Banking & Finance 37 (8), 2779–2792.
32
Kaiser, T. and L. Menkhoff (2017). Does financial education impact financial be-
havior, and if so, when? Technical report, World Bank Policy Research Working
Paper No. 8161.
Kezdi, G. and R. J. Willis (2008). Stock market expectations and portfolio choice
of american households. work in progress.
Kezdi, G. and R. J. Willis (2011). Household stock market beliefs and learning.
Technical report, National Bureau of Economic Research.
Klapper, L., A. Lusardi, and G. A. Panos (2013). Financial literacy and its conse-
quences: Evidence from russia during the financial crisis. Journal of Banking &
Finance 37 (10), 3904–3923.
Koetse, M., R. Florax, and H. de Groot (2013). Correcting for primary study
misspecifications in meta-analysis. Technical report, Tinbergen Institute.
Lipsey, M. W. and D. B. Wilson (2001). Practical meta-analysis. Sage Publications,
Inc.
Luhrmann, M., M. Serra-Garcia, and J. Winter (2015). Teaching teenagers in fi-
nance: Does it work? Journal of Banking & Finance 54, 160–174.
Lusardi, A., P.-C. Michaud, and O. S. Mitchell (2017). Optimal financial knowledge
and wealth inequality. Journal of Political Economy 125 (2), 431–477.
Lusardi, A. and O. S. Mitchell (2008). Planning and financial literacy: How do
women fare? American Economic Review 98 (2), 413–17.
Lusardi, A. and O. S. Mitchell (2011). Financial literacy around the world: An
overview. Journal of Pension Economics & Finance 10 (4), 497–508.
Lusardi, A. and O. S. Mitchell (2014). The economic importance of financial literacy:
Theory and evidence. Journal of Economic Literature 52 (1), 5–44.
Lusardi, A., O. S. Mitchell, and V. Curto (2010). Financial literacy among the
young. Journal of Consumer Affairs 44 (2), 358–380.
Mandell, L. and L. S. Klein (2009). The impact of financial literacy education on sub-
sequent financial behavior. Journal of Financial Counseling and Planning 20 (1).
McCannon, B. C., C. T. Asaad, and M. Wilson (2016). Financial competence,
overconfidence, and trusting investments: Results from an experiment. Journal
of Economics and Finance 40 (3), 590–606.
33
Meier, S. and C. D. Sprenger (2015). Temporal stability of time preferences. Review
of Economics and Statistics 97 (2), 273–286.
Merton, R. C. (1969). Lifetime portfolio selection under uncertainty: The
continuous-time case. The Review of Economics and Statistics , 247–257.
Miller, M., J. Reichelstein, C. Salas, and B. Zia (2015). Can you help someone
become financially capable? A meta-analysis of the literature. The World Bank
Research Observer 30 (2), 220–246.
Mohapatra, S., D. Ratha, and A. Silwal (2011). Outlook for remittance flows 2011-
13: Remittance flows recover to pre-crisis levels.
OECD (2015a). 2015 update of the OECD/INFE toolkit for measuring financial
literacy and financial inclusion.
OECD (2015b). National strategies for financial education: OECD/INFE policy
handbook. Handbook, OECD.
OECD (2017). G20/OECD INFE report on ensuring financial education and con-
sumer protection for all in the digital age. Technical report, OECD.
Rajan, R. G. (2011). Fault lines: How hidden fractures still threaten the world
economy. Princeton University press.
Scholz, J. K., A. Seshadri, and S. Khitatrakun (2006). Are Americans saving “op-
timally” for retirement? Journal of Political Economy 114 (4), 607–643.
Schuhen, M. and S. Schurkmann (2014). Construct validity of financial literacy.
International Review of Economics Education 16, 1–11.
Seshan, G. and D. Yang (2014). Motivating migrants: A field experiment on fi-
nancial decision-making in transnational households. Journal of Development
Economics 108, 119–127.
Skimmyhorn, W. et al. (2016). Assessing financial education: Evidence from boot
camp. American Economic Journal: Economic Policy 8 (2), 322–343.
Van Rooij, M., A. Lusardi, and R. Alessie (2011). Financial literacy and stock
market participation. Journal of Financial Economics 101 (2), 449–472.
Von Gaudecker, H.-M. (2015). How does household portfolio diversification vary
with financial literacy and financial advice? The Journal of Finance 70 (2), 489–
507.
Willis, L. E. (2008). Against financial-literacy education. Iowa L. Rev. 94, 197.
34
Willis, L. E. (2011). The financial education fallacy. American Economic Re-
view 101 (3), 429–34.
Xu, L. and B. Zia (2012). Financial literacy around the world: An overview of the
evidence with practical suggestions for the way forward. Policy Research Working
Paper Series 6107, The World Bank.
35